Jensen Huang exit threatens $3.2 billion in China AI projects

More than 23 billion yuan ($3.2 billion) of Chinese AI computing projects built around Nvidia H100, H200 and H800 chips now face delays, supply disruption or costly redesigns as Washington weighs tighter export controls.

Reuters reviewed more than 150 procurement notices, investment plans, construction contracts and corporate disclosures. The amount is not a single confirmed capacity total. It combines planned investments, signed contracts and completed installations, and covers chips, buildings and power systems.

Most buyers are local government-backed ventures, universities and regional computing centers rather than China’s largest technology companies.

What does Jensen Huang’s departure change?

Reuters reported the risk after Jensen Huang resigned as Nvidia’s chief executive. Huang was Washington’s loudest advocate for easing chip curbs. His exit reduces Nvidia’s lobbying leverage as lawmakers press for tighter limits and the Commerce Department prepares a rule on remote access to chips in overseas data centers.

“Washington must stop focusing solely on who buys the chips on the surface,” said Sunny Cheung, an analyst at the Jamestown Foundation in Washington. He argues that enforcement should track who uses the computing power.

Nvidia and China’s commerce ministry did not respond to Reuters’ requests for comment. Nvidia said in an August filing that it was effectively shut out of China’s data center computing market and excludes China data center revenue from its forecasts.

Huang said last year that Nvidia’s share of China’s market had fallen from 95% to zero. An October IDC report cited by Reuters separately estimated that Nvidia still accounted for 34% of Chinese AI-accelerator shipments in the first half of 2026.

In January, the Trump administration began reviewing H200 export licenses for eligible Chinese customers case by case. The case-by-case process leaves buyers exposed to each new licensing decision and any shift toward tighter enforcement.

Reuters said the identified projects have not been linked to Megaspeed, the Singapore Nvidia cloud partner at the center of allegations involving former executive Alice Huang. The documents do not establish how the restricted hardware reached Chinese users. Megaspeed denied illegally diverting Nvidia chips to China.

Which projects carry the most exposure?

The fact box uses reported values and does not itemize every project included in the 23 billion-yuan total.

• Xianyou county, Fujian: A proposed 6.2 billion-yuan ($873 million) AI computing center would use Nvidia H100 chips. The project still seeks commercial partners.
• Shaanxi: A subsidiary of state-owned engineering group Sinomach won a 3 billion-yuan contract for a computing center using H100 and H200 chips.
• Jilin: An April government prospectus describes a 2.26 billion-yuan project pairing H100s with Chinese-developed chips. It expects bank loans to cover about 80% of the investment.
• Inner Mongolia: A 1.5 billion-yuan development plans 625 Nvidia H200 servers. Construction began on two first-phase buildings in June, according to industry reporting cited by Reuters.
• Gansu: A consortium led by a subsidiary of state-owned Shaanxi Construction Engineering won a 1.16 billion-yuan construction contract in July. Tender documents specify about 3,072 Nvidia H100 and H200 processors. Reuters could not establish whether the chips were delivered.
• Wuxi: A cluster running H800 chips entered operation in March 2025, according to the city government. The cluster had no stated value.
• Fujian: Shanghai-listed Damingcheng disclosed in June that a 759 million-yuan computing-services agreement for an H800 cluster had ended early.
• University purchases: Shenzhen University won a contract for three H100 processors in September. US-blacklisted Northwestern Polytechnical University bought two H200 chips the following month.

The figures cover several commercial stages. Xianyou remains a proposal, Shaanxi and Gansu rest on construction contracts, Wuxi is operating, and the university purchases represent individual transactions rather than regional clusters.

How are Chinese technology shares reacting?

Alibaba fell as much as 7.3% in Hong Kong on Thursday. Z.AI and MiniMax each lost at least 8% as investors considered tighter US scrutiny of cloud operators’ access to advanced chips. Alibaba rents Nvidia chips through Megaspeed.

The selloff continued on Friday morning. The Hang Seng Index fell 1.4%, Z.AI dropped 7.4%, MiniMax 4.9%, Kuaishou 3.1%, Tencent 2.6% and Alibaba 0.6%. The CSI 300 slipped 0.5%.

Julia Pan of UOB Kay Hian said Chinese cloud providers have stockpiled chips since the start of the year. She expects their buffers to last, but expects research timelines to slip as operators wait for clearer access to advanced processors.

Beijing on Thursday urged Washington not to use the spying allegations against Alice Huang to justify new technology restrictions. Alibaba and Tencent report September-quarter results next week.

The next commercial test is whether Chinese buyers can keep existing clusters supplied without costly redesigns or delayed research schedules as Washington tightens access.